Encription--wireless versus browesr

I thought I knew quite a bit about encryption, apparently wrong.

I always run with 64-bit encryption, everything perfect, except 1 site. Get
excellent signal strength, excellent down/uploads with dslreports.

Turns out this 1 site (smithbarney.com) that is extremely slow, or, most of
the time, doesn't even come up, uses SSL, 128-bit https encryption at the
browser level.

As an experiment, I temporally turned encryption off both at the router & at
the client. Now this site runs very fast, as it should. Since I want to run
w/encryption, I tried 128-bit wireless encryption. Has same problem as
64-bit encryption. Switched back to no encryption, 64-bit several times.
Quite consistent, this 1 secure site only runs well with no wireless
encryption.

I've tried other https sites (while wireless encryption was 64-bit), run
fine. Only problem I see is with smithbarney site.

My understanding is the wireless encryption runs just above the physical
layer. And that it totally unrelated to the browser 128-bit encryption, I
would guess at the application layer.

I am also running with MAC address filtering, feel that is not part of the
equation.

I went 1 step further & dialed up to the Internet with this 1 problem
machine. Ran the troubled site just fine, slower, but in the range of
dialup.

Is my understanding correct? Can anyone think of why this 1 site has a
problem? Can anyone think of something I can try to diagnose this, I really
want to run with encryption on?

Norm

(These is actually a 101 in my email address, not 102)

You're right, WEP is usually implemented at the MAC layer in the chipset,
which is just above the physical layer. I've read that old 802.11b equipment
may suffer performance degradation when WEP is enabled, if encryption is
implemented via the host processor.

If your driver is actually doing the WEP payload encryption, then it's doing
an RC4 encryption twice for every packet exchanged with an https site (once
for https, and once for WEP). That means it runs a pseudorandom number
generator to calculate a string of bytes the same size as the message, and
then exclusively ors the message - and it does that twice for every message.

I suppose that if the SmithBarney site exchanges a lot more data than the
other https sites you visit (are they just userid/password prompts?), it
might push you over the edge for processor power. This is only plausible if
you're using older 802.11 equipment and a not-too-fast host processor.

"Captain Norm" <(E-Mail Removed)> wrote in message
news:aYwwb.23012$(E-Mail Removed) et...
> I thought I knew quite a bit about encryption, apparently wrong.
>
> I always run with 64-bit encryption, everything perfect, except 1 site.
Get
> excellent signal strength, excellent down/uploads with dslreports.
>
> Turns out this 1 site (smithbarney.com) that is extremely slow, or, most
of
> the time, doesn't even come up, uses SSL, 128-bit https encryption at the
> browser level.
>
> As an experiment, I temporally turned encryption off both at the router &
at
> the client. Now this site runs very fast, as it should. Since I want to
run
> w/encryption, I tried 128-bit wireless encryption. Has same problem as
> 64-bit encryption. Switched back to no encryption, 64-bit several times.
> Quite consistent, this 1 secure site only runs well with no wireless
> encryption.
>
> I've tried other https sites (while wireless encryption was 64-bit), run
> fine. Only problem I see is with smithbarney site.
>
> My understanding is the wireless encryption runs just above the physical
> layer. And that it totally unrelated to the browser 128-bit encryption, I
> would guess at the application layer.
>
> I am also running with MAC address filtering, feel that is not part of the
> equation.
>
> I went 1 step further & dialed up to the Internet with this 1 problem
> machine. Ran the troubled site just fine, slower, but in the range of
> dialup.
>
> Is my understanding correct? Can anyone think of why this 1 site has a
> problem? Can anyone think of something I can try to diagnose this, I
really
> want to run with encryption on?
>
> Norm
>
> (These is actually a 101 in my email address, not 102)
>
>